Research Question: can migration and residency behaviour be inferred from the depth log?

vertical space use

Depth median, min and max

female
male

There are peaks in the change of daily median in September, December (this is influenced by the non-accurate depth measurements, though), March and May.
There seems to be no clearly visible pattern in daily vertical migration, comparing between potentially migrating and residential periods.
These months I will have a further look into.

Here we see peaks in the change of daily median in October 2018, May 2019, August 2019 and October 2019.
Both in October 2018 and 2019, we see reverse Diel Vertical Migration.

Note: a rolling mean over 3 days was applied to smooth the data.

Daily Vertical Migration

Diel Vertical Migration (DVM): Depth during day significantly deeper than during night.
reverse DVM (rDVM): Depth during night significantly deeper than during day.
no DVM (nDVM): No significant difference between depth during night and day.

There does not seem to be a pattern between potential activity (migrating/residing) and Daily Vertical Migration behaviour.

female
male

Vertical speed vs depth

Vertical speed = the absolute difference between two subsequent measurements divided by the number of samples taken per minute (Units = m/min).

There is a bimodal distribution in monthly vertical speed, no matter the depth that was swum in and the potential activity (migrating/resting).

female
male

Potential Migration months

female
male

Sep-Oct

The interesting period is between Sep 20 and October 15.

Oct-Nov

Horizontal distance

We need to be careful interpreting this since it’s a model output.

Overall, the female shark (tag 308) travelled 1462.54 km and the male shark (tag 321) travelled 1272.07 km.
The horizontal distance covered by the female shark (tag 308) is 536.52 km, and for the male shark (tag 321) it is 173.93.

You can select a zoom window, and select/deselect data by clicking on then on the legend.

the largest horizontal distance travelled per day is roughly 30 km.

k means clustering

Can unsupervised learning distinguish migrating and resident behaviour?

PCA on daily dst summary statistics (median depth, depth range, vertical speed, …), retain the forst 2 PC axes.
the k means clustering I did can distinguish between summer and winter residential behaviour, but is not able to differentiate the migrating periods in between.

female
male

Map

--- title: "DST migration analyses" description: "When do tag 308 and 321 start/stop migrating?" # author: "Lotte Pohl" execute: echo: false include: true output: html_document: df_print: paged theme: readable # toc_float: true toc: true # number_sections: false # toc_depth: 2 date: 3.20.2023 editor_options: chunk_output_type: console knitr: opts_chunk: collapse: true format: html: page-layout: article code-tools: true fig-width: 8 fig-height: 5 toc: true image: image.PNG draft: false --- ```{r libraries} #| include: false library(tidyverse) library(dplyr) library(plotly) library(ggplot2) library(sf) library(suncalc) library(dplyr) library(lubridate) library(pracma) library(psdr) library(StreamMetabolism) ``` ```{r workspace} #| include: false # rm(list = ls()) dir_path <- "C:/Users/lotte.pohl/Documents/github_repos/ADST_Mustelus_asterias" # source(paste0(dir_path, "/02_scripts/02_load_data/load_depth_temp_logs.R")) # source(paste0(dir_path, "/02_scripts/04_analyses/FFT/calculate_fft_psd.R")) # source(paste0(dir_path, "/02_scripts/05_plots_maps/plot_fft_periodogram.R")) # source(paste0(dir_path, "/02_scripts/05_plots_maps/plot_dst_pressure_temp.R")) # source(paste0(dir_path, "/02_scripts/05_plots_maps/plots_dst_pca_kmeans.R")) source(paste0(dir_path, "/02_scripts/05_plots_maps/plots_dst_summary.R")) source(paste0(dir_path, "/02_scripts/05_plots_maps/plots_dst_pca_kmeans.R")) source(paste0(dir_path, "/02_scripts/05_plots_maps/base_map.R")) source(paste0(dir_path, "/02_scripts/02_load_data/load_dst_geolocation_output.R")) ```  > Research Question: can migration and residency behaviour be inferred from the depth log? # vertical space use   ## Depth median, min and max :::{.panel-tabset} ### female ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 308. #| - "" p_308_depth_median_range_change_ribbon %>% plotly::ggplotly() %>% plotly::layout(legend = list(orientation = "h", x = 0.5, y = -0.2, xanchor = "center", yanchor = "top")) ``` * There are peaks in the change of daily median in September, December (this is influenced by the non-accurate depth measurements, though), March and May. * There seems to be no clearly visible pattern in daily vertical migration, comparing between potentially migrating and residential periods. * These months I will have a further look into.  ### male ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" p_321_depth_median_range_change_ribbon %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.2, xanchor = "center", yanchor = "top")) ``` * Here we see peaks in the change of daily median in October 2018, May 2019, August 2019 and October 2019. * Both in October 2018 and 2019, we see reverse Diel Vertical Migration. ::: *Note: a rolling mean over 3 days was applied to smooth the data.* ## Daily Vertical Migration * Diel Vertical Migration (DVM): Depth during day significantly deeper than during night. * reverse DVM (rDVM): Depth during night significantly deeper than during day. * no DVM (nDVM): No significant difference between depth during night and day. There does not seem to be a pattern between potential activity (migrating/residing) and Daily Vertical Migration behaviour. :::{.panel-tabset} ### female  ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 308. #| - "" p_tag308_DVM_weekly %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` ### male ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" p_tag321_DVM_weekly %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` ::: ## Vertical speed vs depth Vertical speed = the absolute difference between two subsequent measurements divided by the number of samples taken per minute (Units = m/min). * There is a bimodal distribution in monthly vertical speed, no matter the depth that was swum in and the potential activity (migrating/resting). :::{.panel-tabset} ### female ```{r} #| layout-ncol: 2 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 308. #| - "" # p_308_vertical_speed %>% plotly::ggplotly() %>% # layout(legend = list(orientation = "h", # x = 0.5, y = -0.5, # xanchor = "center", yanchor = "top")) p_308_vertical_speed_splitviolin p_308_depth_month_splitviolin ``` ### male ```{r} #| layout-ncol: 2 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" # p_321_vertical_speed %>% plotly::ggplotly() %>% # layout(legend = list(orientation = "h", # x = 0.5, y = -0.5, # xanchor = "center", yanchor = "top")) p_321_vertical_speed_splitviolin p_321_depth_month_splitviolin ``` ::: # Potential Migration months :::{.panel-tabset} ### female  #### Sep-Oct The interesting period is between Sep 20 and October 15. ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 308. #| - "" p_308_mig1_depth_DVM %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ```  ### male #### Oct-Nov ```{r} #| layout-ncol: 2 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" p_321_mig1_depth_DVM %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) p_321_mig2_depth_DVM %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` ::: # Horizontal distance **We need to be careful interpreting this since it's a model output.** * Overall, the female shark (tag 308) travelled `r tag_308_totalhordistance %>% round(digits = 2)` km and the male shark (tag 321) travelled `r tag_321_totalhordistance %>% round(digits = 2)` km. * The horizontal distance covered by the female shark (tag 308) is `r total_horizontal_distance %>% filter(tag_serial_number == "1293308") %>% dplyr::select(total_horizontal_distance_km) %>% pull() %>% round(digits = 2)` km, and for the male shark (tag 321) it is `r total_horizontal_distance %>% filter(tag_serial_number == "1293321") %>% dplyr::select(total_horizontal_distance_km) %>% pull() %>% round(digits = 2)`. *You can select a zoom window, and select/deselect data by clicking on then on the legend.* ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" p_308_321_hordistance %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` * the largest horizontal distance travelled per day is roughly 30 km. # k means clustering > Can unsupervised learning distinguish migrating and resident behaviour? * PCA on daily dst summary statistics (median depth, depth range, vertical speed, ...), retain the forst 2 PC axes. * the k means clustering I did can distinguish between summer and winter residential behaviour, but is not able to differentiate the migrating periods in between. :::{.panel-tabset} ## female ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" plot_f_308_dailydepthmedian_cluster %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` ## male ```{r} #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-cap: Ribbon plot showing the daily median, minimum and maximum depth of tag 321. #| - "" plot_m_321_dailydepthmedian_cluster %>% plotly::ggplotly() %>% layout(legend = list(orientation = "h", x = 0.5, y = -0.5, xanchor = "center", yanchor = "top")) ``` ::: # Map ::: {.column-screen-inset} ```{r} #| include: true #| layout-ncol: 1 ##| fig-height: 4 #| column: page #| fig-subcap: #| - "" # for the long term DST tracks pal_dst <- colorNumeric(palette = "magma", domain = c(masterias_dst_geolocation_output$date_time %>% min(), masterias_dst_geolocation_output$date_time %>% max())) # mpts_all$date_time, masterias_detections_clean$date_time # mpts_short <- mpts_all %>% filter(!tag_serial_number == c("1293308", "1293321")) legend_dates = c("2018-08-01", "2018-11-01", "2019-02-01", "2019-05-01", "2019-08-01", "2019-11-01") base_map %>% hideGroup(c("stations", "<span style=color:grey>released tags</span>", "areas")) %>% # add tag 308 #### addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293308"), lat = ~detection_latitude, lng = ~detection_longitude, color = "#B5B5B5", weight = 3, opacity = 0.6, label = ~tag_serial_number, group = "tag 308 (female)") %>% addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293308"), lat = ~detection_latitude_mean, lng = ~detection_longitude_mean, color = "#B5B5B5", weight = 1, fillOpacity = 0.2, opacity = 0.2, dashArray = "5,5", label = ~paste0(tag_serial_number, ", mean track"), group = "mean tracks") %>% addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293308"), lat = ~detection_latitude_mode, lng = ~detection_longitude_mode, color = "#B5B5B5", weight = 1, fillOpacity = 0.3, dashArray = "1,7", label = ~paste0(tag_serial_number, ", mode track"), group = "mode tracks") %>% addCircleMarkers(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293308"), lat = ~detection_latitude, lng = ~detection_longitude, fillColor = ~pal_dst(date_time), radius = 5, weight = 0, fillOpacity = 0.8, label = ~paste0(format(date_time,"%F"), ", ", sex), group = "tag 308 (female)") %>% # add tag 321 #### addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293321"), lat = ~detection_latitude, lng = ~detection_longitude, color = "#B5B5B5", weight = 3, opacity = 0.6, label = ~tag_serial_number, group = "tag 321 (male)") %>% addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293321"), lat = ~detection_latitude_mean, lng = ~detection_longitude_mean, color = "#B5B5B5", weight = 1, fillOpacity = 0.2, opacity = 0.2, dashArray = "5,5", label = ~paste0(tag_serial_number, ", mean track"), group = "mean tracks") %>% addPolylines(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293321"), lat = ~detection_latitude_mode, lng = ~detection_longitude_mode, color = "#B5B5B5", weight = 1, fillOpacity = 0.3, dashArray = "1,7", label = ~paste0(tag_serial_number, ", mode track"), group = "mode tracks") %>% addCircleMarkers(data = masterias_dst_geolocation_output %>% filter(tag_serial_number == "1293321"), lat = ~detection_latitude, lng = ~detection_longitude, fillColor = ~pal_dst(date_time), radius = 5, weight = 0, fillOpacity = 0.8, label = ~paste0(format(date_time,"%F"), ", ", sex), group = "tag 321 (male)") %>% # ADD CONTROLS ##### leafem::addMouseCoordinates() %>% addLayersControl(position = "topright", baseGroups = c("EMODnet bathymetry", "satellite"), overlayGroups = c("tag 308 (female)", "tag 321 (male)","mean tracks", "mode tracks", "wrecks, OWFs, cables"), options = layersControlOptions(collapsed = FALSE)) %>% hideGroup(c("mean tracks", "mode tracks", "wrecks, OWFs, cables")) %>% addLegend(position = "topleft", colors = legend_dates %>% as.POSIXct() %>% pal_dst(), labels = legend_dates, opacity = 1, title = "Dates") ``` :::